A 2-D framework for accelerating the development of physics schemes in high resolution global modeling

Abstract:

The horizontal resolution of global climate models has increased dramatically in recent years. The increase in resolution can improve not only the numerical solution of the dycore, but also the behavior of physics schemes in the model. However, an increase in resolution carries an enormous increase in the cost of computations and amount of output. This becomes more serious in the model development stage because numerous tests are required to assess the performance of new schemes, and the performance should be assessed in a similar domain and a similar resolution to the target simulations. We approach this problem by applying 2-D Hadley circulation modeling that simulates many types of clouds, including low-level clouds and scale interactions. These cannot be simulated by a single column model or coarse-grid global model. By neglecting the zonal extension, we can save computation, and easily increase model resolution and the number of test cases. We have imported and are assessing the new physics schemes planned to be used in a future global system resolving version of DOE's Energy Exascale Earth System Model (E3SM). The target resolutions are 16 km, 8 km, and 4km, and we have also performed 2-D Hadley circulations at LES resolution (250 m) as a reference. In a simulation using 8 km horizontal grid spacing with 128 vertical levels in a domain from the north-pole to the south-pole, a three-year integration can be carried out with 300 node-hours on a supercomputer. The results show realistic features of the Hadley circulation. In a higher resolution simulation, the simulated Hadley circulation width tends to be narrower and radiative cooling in the lower troposphere in equatorial regions becomes more marked. These suggest that the 2-D framework can serve as a testbed for assessing, inter alia, model parameterizations and their influence on the circulation. Lastly, implementation of a parallelized FIVE (Framework for Improvement by Vertical Enhancement; Yamaguchi et al. 2017) into this 2-D framework is being applied to improve the vertical resolution and the computational efficiency. We expect this 2-D framework to rapidly accelerate the development of physics schemes in climate models.